Covid-19 Deaths and County Health Risk Factors — Population Density and Distance to Supermarkets

Pete Zajonc
10 min readJun 5, 2020



This US County health analysis is dedicated to individuals who are the victims of the Covid-19 virus and to all the people working hard to keep our US Counties safe and nourished.


The risk of covid-19 death is not equitable and increases with population density. Someone who is contagious with the coronavirus walking in a crowd is a risk to themselves and others. Like someone with a loaded gun or a teenager driving in a crowd, the actuarial risks associated with someone who is infectious are frighteningly high. 80% of the time there is no problem — the other 20% of the time is a disaster. Family and friends may die, the carrier, too. And we don’t know which 20% is most dangerous. It might be best if everyone gets tested, and best if everyone gets licensed.

Population density is a risk factor because it effects community spread of the covid-19 disease. Since infections come from sustained close contact they occur most easily when people cluster together. From this perspective Nevada should be considered densely populated with a high potential health risk. The US military used great swaths of Nevada to test atom bombs, but Nevada’s cities are large and tightly packed and risk community transmission of the virus. And particularly for Las Vegas, tourists bring their gambling dollars to the city, and the city is gambling with virus infections that tourists may spread. What happens in Vegas…

To capture this aspect of clustered population densities this analysis takes advantage of US Department of Agriculture data on County populations that must travel under (or over) 1 mile in order to visit a supermarket. A highly clustered population — living within 1 mile of a supermarket — is also one that lives close to restaurants, community centers, hospitals, churches, and nursing homes. All other things being equal, the larger the share of residents clustered near supermarkets, the larger the risk of a covid-19 infection.

This analysis get into a more County details later, but for now the chart below illustrates how supermarket distance serves as a clustering yardstick of state population density and its association with covid-19 death rates. The upward sloping trendline indicates that densely populated states, like New York (NY), have both high density and high covid-19 death rates — the numbers of covid-19 deaths taken as a percentage of state population. There are exceptions: California (CA) and Nevada (NV) in the upper-left of the chart are densely populated but have low death rates.

Population Density of States and Covid-19 Death Rates

The chart indicates that public health risk from covid-19 is not only explained by population density but also by each state’s public health response along with other health risk factors (see other analyses in the this series for more). Consider the series of events that occurs with each case: exposure, incubation, infection, infection of others, testing, hospitalization, and final outcome. State and local interventions, such as social distancing, testing, contact tracing, and quarantining attempt to limit the risks of each step along the way in order to improve overall covid-19 public health outcomes.

If nothing were done to combat the virus by public health officials — as some proposed early on, and only ended up delaying these essential, life-saving US public health interventions — then population density by itself would explain more transmissions, infections and deaths. Though more should have been done sooner, the testing, quarantining, and general healthcare that was initiated avoided even more hospitalization and covid-19 deaths. It should have been better, but it could have been worse.

High testing rates are also associated with high population density as the upward trendline in the next chart shows. Testing interventions is exemplified by Rhode Island (RI) — at the virus’s US epicenter — having tested over 15% of its population. States to the right of the chart are all in US’s covid epicenter — densely populated and having high testing rates. But, some high density states do not report high testing rates — California (CA) and Nevada (NV). It is likely California and Nevada used other public health measures (“lockdowns”) to keep their early infection rates. Meanwhile, Las Vegas is very cautiously rolling the dice to welcome back patrons on June 4.

Population Density of States and Percent Tested

The last chart in this section shows the association between state testing and covid-19 death rates. As a rule, when testing is done aggressively, it is associated with lower death rates as reflected in the upward sloping trendline. Rhode Island’s aggressive testing strategy is associated with dramatically low covid-19 death rates. Along its other public health mandates, Rhode Island testing undoubtedly kept it safer than other New England states.

Another use of the chart’s trendline is to indicate to state public health officials where more testing may help. Pennsylvania (PA), Michigan (MI), Maryland (MD), Indiana (IN), and Ohio (OH) all have high death rates from covid-19 but relatively low testing rates. As shown in the previous two charts, most of these state have relatively high population densities.

Percent of State Population Tested and Covid-19 Death Rates

Some may argue that this observational data indicates, for example, that Rhode Island tested more than necessary. Ultimately, especially where population density is high, it seems that erring on the side of too many tests and public health interventions is most likely to put the country back onto the quickest road to a sustainable health and economic recovery.

The analysis below of US Counties provides more detail and insight into population density as a health risk factor for covid-19 outcomes.


Community transmission occurs when residents are in close contact with each other and naturally increases with high density populations clustered together. These conditions also lead County public health officials to make interventions, including social distancing, testing and contact tracing, to combat the health risks of high population density. In the largest US Counties the struggle between ongoing public health efforts and covid-19 will be with us for months to come. However, with the safety that comes from living in sparsely populated Counties comes a risk that when covid-19 disease strikes it is more often fatal, perhaps because of poor hospital access.

Analysis of Covid-19 Deaths — Data as of June 2, 2020

33% of the US population live in sparsely populated, low density Counties where half or more residents must travel over 1 mile to shop at a supermarket for food (highlighted in “orange” bars in the chart below). None of these are large 1 million and over “A” Counties. Of concern, the most sparsely populated, rural “D” Counties have 30% higher death rates than the rest of “D” Counties — 1.3 times more deaths per 10,000 people occur in low density “D” Counties than high density “D” Counties where populations are more clustered. As of today, rural “D” Counties have only 2% of all US covid-19 deaths, so perhaps these 2,309 deaths are less of a concern but this is more than the entire country of Switzerland and may be instructive to understand why death rates are lower in densely populated “D” Counties and higher in sparsely populated “D” Counties.

Sparse Populations and Covid-19 Deaths for Counties where 50% or more residents must travel 1 mile or more to get to a supermarket (orange)

Highly dense populations, such as those in “A” and “B” Counties, have the most Covid-19 deaths and higher deaths per 10,000 residents than the rest of the US. County public health official aggressive testing interventions designed to minimize these risks, and may be paying off. In fact, for all covid-19 cases reported death rates are lower in the most densely populated Counties, and the least densely populated Counties are struggling to saving lives when cases are identified. Most likely, in the rare cases where infections occur and travel distances for hospital care is long, the prospects of surviving the infection decrease.

As the chart below illustrates, when cases are reported in sparsely populated Counties there is an increased chance of death for each case (“orange” bars). Counties with the most remote population have poorer success saving lives among new covid-19 cases. For example, as the table shows, 50% more residents per 10,000 who test positive for covid-19 in sparsely populated “C” Counties will die compared to other “C” Counties that have more clustered, dense populations. While they have fewer cases overall, the disadvantage of rural populations is that they do not have nearby medical facilities and public health services. Once a case is confirmed, Counties with more dense populations (“blue” bars) — and, presumably, better hospital access — report fewer deaths.

Last 4 Weeks Deaths as a Percentage of New Covid-19 Cases for Counties that have Sparse Populations where 50% or more residents must travel 1 mile or more to get to a supermarket (orange)

Over the past 4 weeks, approximately 20% of positive covid-19 tests and deaths occur in sparsely populated, less dense US Counties. Public health programs and hospitals in sparsely populated Counties are struggling to save the lives of covid-19 patients and must respond with better hospitalization and care.

Counties of all sizes are vulnerable to the virus. Public health programs should prepare residents to be proactive because the onset of covid-19 symptoms requiring medical care happens in days. If food shopping in your County can leave you hungry, there’s a fair bet County hospital services will leave you hurting.

Analysis Approach

This is one of a series of analyses that relate covid-19 deaths to key County health and demographic characteristics, including risk factors such as cancer death rates or percent of adults over 60. Each analysis focuses on the Counties’ risk factors that are higher than some “average”. The “average” is defined here as the characteristic’s incidence or percentage — for example, average cancer death rate or percentage of adults over 60 — and this risk group of Counties accounts for roughly 1/3rd of the US population when pooled together. Covid-19 deaths are then totaled for Counties with and without the risk factor.

Side-by-side bar charts illustrate how Counties are faring in the covid-19 crises when grouped by size of population and median income and split by above and below some average for a risk factor.

Covid-19 death rates for resulting 1/3 versus 2/3 population comparisons are then written as a ratio, such as, 0.5 : 1. Of course, the death rate ratios change as the data changes. For example, a 0.8 : 1 ratio compares the covid-19 death rates for Counties with and without the above average percent adults over age 60 risk factor. All things being equal, a 1.0 : 1 ratio means the covid-19 death rates match the ratio of the County populations. Typically it does not.

Here’s an example. The rate of covid-19 deaths is deaths divided by population. It’s 0.03% today, meaning 3 out of 10,000 US residents have died due to covid-19. In these studies we compare death rates for groups of Counties with and without the risk factors in a covid-19 death rate ratio. It is an “odds” ratio of death occurring with and without the risk factor. The 1.2 : 1 ratio below is a comparison of the death rates between “A” Counties with and without above adult age 60 risk factor. The ratio of the death rates is 1.2 = (3,522 /7,585,831) / (33,043/ 85,069,756). It tells us that the odds of dying from covid-19 are 20% higher for populations in large “A” Counties with above average adult age 60 populations.

Example: “A” Counties where Adult Age 60+ is 23% or more of County Population

Want More on Covid and County Health Risks?

Respiratory DiseaseHow are small, rural Counties with high rates of COPD surviving the ravages of covid-19?

Presidential Pary Voting Party — What does loyal presidential party voting deliver to Counties in the time of covid-19?

Diabetes Incidence — How are Counties with high Diabetes rates faring during the crisis?

Above Average White, Hispanic, and African American Counties — All men and women are created equal. What could make the Counties they live in more equitable?

Data Sources

US Census, USA Facts Covid-19, IHME forecasts, US Cancer Deaths NIH 5 year average, CDC Handbook on Death Reporting, 2003; New York Times; Median Income; CDC Wonder Detailed Mortality; CDC Fine Particulate Matter 2003–2011; MIT Election Data + Science Lab; Michigan Monthly Deaths; Testing from Covid Tracking Project

Originally published at



Pete Zajonc

Experienced analytic consultant at Epsilon and Time Inc. His mission is to create value and strategic insight from data analytics.